September 8, 2011 § Leave a comment
Got the Kinect working on my Mac (Snow Leopard) under two different methods.
1. Using Shiffman’s library, and using raw depth data for point tracking. This is the simplest to install and code for but limited in terms of the range of user input.
2. Using OSCeleton, a skeleton of users could be generated. This is definitely the more flexible and precise of the two options (code setup also took quite long, and could be potentially more complex). However, there is quite a long delay for this system to callibrate with a user’s skeleton.
[ Video ]
Therefore, OSCeleton is great for games/interfaces that engages players for long periods of time and requires precise gesture tracking. Using raw data, such as Shiffman’s, would be more appropriate for temporal experiences.
I have decided to work with the raw depth data, potentially extending it later on with OpenCV, and even (in the future) integrating skeleton tracking to achieve the benefits of both methods.
- Shiffman’s Kinect: http://www.shiffman.net/p5/kinect/
- Instructions for using and setting up OSCeleton:
• Step 1: https://github.com/avin2/SensorKinect
• (Step 1 with more detail: http://tohmjudson.com/?p=30 )
• Additional Step: http://groups.google.com/group/osceleton/browse_thread/thread/59b2f616cece788?pli=1